Proof-of-concept RegExp-cache for exec, test, replace.

src/regexp-delay.js

Index: src/regexp-delay.js
diff --git a/src/regexp-delay.js b/src/regexp-delay.js
index 7bec455d37fedd46702d36f02db20ed6d6396d71..606ca17ffd9e6d5d4739667b4dc4f09029d19f6e 100644
--- a/src/regexp-delay.js
+++ b/src/regexp-delay.js
@@ -140,22 +140,43 @@ function DoRegExpExec(regexp, string, index) {
 }
 
 
+var regExpExecCacheResult;
+var regExpExecCacheInput;
+var regExpExecCacheId;
+var regExpExecCacheIndex;
+
 function RegExpExec(string) {
   if (!IS_REGEXP(this)) {
     throw MakeTypeError('method_called_on_incompatible',
                         ['RegExp.prototype.exec', this]);
   }
   if (%_ArgumentsLength() == 0) {
-    var regExpInput = LAST_INPUT(lastMatchInfo);
-    if (IS_UNDEFINED(regExpInput)) {
+    string = LAST_INPUT(lastMatchInfo);
+    if (IS_UNDEFINED(string)) {
       throw MakeError('no_input_to_regexp', [this]);
     }
-    string = regExpInput;
   }
-  var s = ToString(string);
-  var length = s.length;
+
   var lastIndex = this.lastIndex;
+  if (%_IsIdentical(regExpExecCacheInput, string) &&
+      %_IsIdentical(regExpExecCacheId, %_RegExpId(this)) &&
+      %_IsIdentical(regExpExecCacheIndex, lastIndex)) {
+    var cachedResult = regExpExecCacheResult;  // Cached result.
+    if (cachedResult) {
+      // If non-null, don't return the same array as last time.
+      var newArray = ArraySlice.call(cachedResult, 0, cachedResult.length);
+      newArray.input = cachedResult.input; 
+      newArray.index = cachedResult.index;
+      cachedResult = newArray;
+    }
+    %_Log("regexp", "regexp-exec-cache-hit", []);
+    return cachedResult;
+  }
+  %_Log("regexp", "regexp-exec-cache-miss", []);
+
   var i = this.global ? TO_INTEGER(lastIndex) : 0;
+  var s = ToString(string);
+  var length = s.length;
 
   if (i < 0 || i > s.length) {
     this.lastIndex = 0;
@@ -166,33 +187,41 @@ function RegExpExec(string) {
   // matchIndices is either null or the lastMatchInfo array.
   var matchIndices = %_RegExpExec(this, s, i, lastMatchInfo);
 
+  var result;
   if (matchIndices == null) {
     if (this.global) this.lastIndex = 0;
-    return matchIndices; // no match
-  }
-
-  var numResults = NUMBER_OF_CAPTURES(lastMatchInfo) >> 1;
-  var result = new $Array(numResults);
-  for (var i = 0; i < numResults; i++) {
-    var matchStart = lastMatchInfo[CAPTURE(i << 1)];
-    var matchEnd = lastMatchInfo[CAPTURE((i << 1) + 1)];
-    if (matchStart != -1 && matchEnd != -1) {
-      result[i] = SubString(s, matchStart, matchEnd);
-    } else {
-      // Make sure the element is present. Avoid reading the undefined
-      // property from the global object since this may change.
-      result[i] = void 0;
+    result = matchIndices; // no match
+  } else {
+    var numResults = NUMBER_OF_CAPTURES(lastMatchInfo) >> 1;
+    result = new $Array(numResults);
+    for (var i = 0; i < numResults; i++) {
+      var matchStart = lastMatchInfo[CAPTURE(i << 1)];
+      var matchEnd = lastMatchInfo[CAPTURE((i << 1) + 1)];
+      if (matchStart != -1 && matchEnd != -1) {
+        result[i] = SubString(s, matchStart, matchEnd);
+      } else {
+        // Make sure the element is present. Avoid reading the undefined
+        // property from the global object since this may change.
+        result[i] = void 0;
+      }
     }
+    if (this.global) this.lastIndex = lastMatchInfo[CAPTURE1];
+    result.index = lastMatchInfo[CAPTURE0];
+    result.input = s;
   }
-
-  if (this.global)
-    this.lastIndex = lastMatchInfo[CAPTURE1];
-  result.index = lastMatchInfo[CAPTURE0];
-  result.input = s;
+  regExpExecCacheResult = result;
+  regExpExecCacheInput = string;
+  regExpExecCacheId = %_RegExpId(this);  // May have changed.
+  regExpExecCacheIndex = lastIndex;
   return result;
 }
 
 
+var regExpTestCacheResult;
+var regExpTestCacheId;
+var regExpTestCacheIndex;
+var regExpTestCacheInput;
+
 // Section 15.10.6.3 doesn't actually make sense, but the intention seems to be
 // that test is defined in terms of String.prototype.exec. However, it probably
 // means the original value of String.prototype.exec, which is what everybody
@@ -203,33 +232,47 @@ function RegExpTest(string) {
                         ['RegExp.prototype.test', this]);
   }
   if (%_ArgumentsLength() == 0) {
-    var regExpInput = LAST_INPUT(lastMatchInfo);
-    if (IS_UNDEFINED(regExpInput)) {
+    string = LAST_INPUT(lastMatchInfo);
+    if (IS_UNDEFINED(string)) {
       throw MakeError('no_input_to_regexp', [this]);
     }
-    string = regExpInput;
   }
-  var s = ToString(string);
-  var length = s.length;
+
   var lastIndex = this.lastIndex;
+  if (%_IsIdentical(regExpTestCacheInput, string) &&
+      %_IsIdentical(regExpTestCacheId, %_RegExpId(this)) &&
+      %_IsIdentical(regExpTestCacheIndex, lastIndex)) {
+    %_Log("regexp", "regexp-test-cache-hit", []);
+    return regExpTestCacheResult;  // Always true or false.
+  }
+  %_Log("regexp", "regexp-test-cache-miss", []);
+
   var i = this.global ? TO_INTEGER(lastIndex) : 0;
+  var s = ToString(string);
+  var length = s.length;
 
+  var result;
   if (i < 0 || i > s.length) {
     this.lastIndex = 0;
-    return false;
-  }
-
-  %_Log('regexp', 'regexp-exec,%0r,%1S,%2i', [this, s, lastIndex]);
-  // matchIndices is either null or the lastMatchInfo array.
-  var matchIndices = %_RegExpExec(this, s, i, lastMatchInfo);
+    result = false;
+  } else {
+    %_Log('regexp', 'regexp-exec,%0r,%1S,%2i', [this, s, lastIndex]);
+    // matchIndices is either null or the lastMatchInfo array.
+    var matchIndices = %_RegExpExec(this, s, i, lastMatchInfo);
 
-  if (matchIndices == null) {
-    if (this.global) this.lastIndex = 0;
-    return false;
+    if (matchIndices == null) {
+      if (this.global) this.lastIndex = 0;
+      result = false;
+    } else {
+    if (this.global) this.lastIndex = lastMatchInfo[CAPTURE1];
+    result = true;
+    }
   }
-
-  if (this.global) this.lastIndex = lastMatchInfo[CAPTURE1];
-  return true;
+  regExpTestCacheResult = result;
+  regExpTestCacheInput = string;
+  regExpTestCacheId = %_RegExpId(this);
+  regExpTestCacheIndex = lastIndex;
+  return result;
 }
 
 
@@ -340,7 +383,7 @@ function SetupRegExp() {
   %FunctionSetLength($RegExp.prototype.compile, 1);
 
   // The properties input, $input, and $_ are aliases for each other.  When this
-  // value is set the value it is set to is coerced to a string. 
+  // value is set the value it is set to is coerced to a string.
   // Getter and setter for the input.
   function RegExpGetInput() {
     var regExpInput = LAST_INPUT(lastMatchInfo);